An engineered glutamate in biosynthetic models of heme-copper oxidases drives complete product selectivity by tuning the hydrogen bonding network

Abstract

Efficiently carrying out the oxygen reduction reaction (ORR) is critical for many applications in biology and chemistry, such as bioenergetics and fuel cells, respectively. In biology, this reaction is carried out by large, transmembrane oxidases such as heme-copper oxidases (HCOs) and cytochrome bd oxidases. Common to these oxidases is the presence of a glutamate residue next to the active site, but its precise role in regulating the oxidase activity remains to be understood. In order to gain insight into the role, we herein report that incorporation of glutamate next to a designed heme-copper center in two biosynthetic models of HCOs improves O₂-binding affinity, facilitates protonation of reaction intermediates, and eliminates release of reactive oxygen species. High-resolution crystal structures of the models revealed extended, water-mediated hydrogen bonding networks involving the glutamate. Electron paramagnetic resonance of the cryoreduced oxy-ferrous centers at cryogenic temperature followed by thermal annealing allowed observation of the key hydroperoxo intermediate that can be attributed to the hydrogen bonding network. By demonstrating these important roles of glutamate in oxygen reduction biochemistry, this work offers deeper insights into its role in native oxidases, may guide the design of more efficient artificial ORR enzymes or catalysts for applications such as fuel cell cathodes.

Graphical Abstract

The oxygen reduction reaction (ORR) catalyzed by terminal oxidases, such as heme-copper oxidases (HCOs)^1,2 and cytochrome bd oxidases,³ plays an important role in bioenergetic processes such as respiration. Both HCOs and cytochrome bd oxidases are heme-containing transmembrane proteins that efficiently deliver electrons to oxygen and use the energy of reduction to generate a proton gradient that drives energy-intensive processes such as the synthesis of ATP. In HCOs, the ORR is catalyzed by a heme-copper site, while the reaction is proposed to be carried out by a single heme d center in cytochrome bd oxidases.

In addition to electron delivery, efficient reduction of oxygen by terminal oxidases requires coordinated delivery of protons during the catalytical cycle (Scheme 1). Protons make their way into the active sites by proton pathways that are extended, hydrogen-bonded chains of amino acids and water molecules.⁴ One example is the D-pathway in HCOs which delivers the majority of the protons used in the catalytic cycle and ends at a glutamate (Glu) residue near the active site.⁵ Similarly, crystal and cryo-electron microscopy structures of cytochrome bd oxidase and mutational studies suggest that a Glu near the heme d is critical in oxidase activity.^6–11

Scheme 1.

Proposed HCO reaction mechanism

While biology has evolved through millions of years to produce these oxidases for such important function in bioenergetics, the precise role of the Glu in regulating the oxidase activity remains to be understood. It has been difficult to pinpoint subtle structural features around the Glu, including the weak hydrogen bonding interactions in these large membrane enzymes, for which high-resolution crystal structures are not easily obtained. To overcome this limitation, synthetic models of HCOs using organic molecules have been obtained,¹² but it has been difficult to incorporate residues such as Glu that often play a role in weak, non-covalent interactions at precise location in the models. To address these issues, we have been using small stable proteins with rigid scaffolds, such as sperm whale myoglobin (swMb), to prepare biosynthetic models of HCOs, allowing both active site structures such as the heme-copper center and residues that are involved in weak and non-covalent interactions to be readily incorporated and studied. We have successfully engineered several variants of swMb to be structural and functional mimics of terminal oxidases.^13–18 The initial variant, L29H/F43H-swMb, called Cu_BMb, displayed relatively slow ORR.^19,20 A subsequent improvement was achieved by introducing a tyrosine to yield either F33Y-Cu_BMb or G65YCu_BMb, which achieved higher rates and over 1000 turnovers of oxygen reduction.²⁰ When combined with an efficient redox partner for electron transfer, one of the variants displayed ORR rate that rivals that of native HCOs.^21,22 Although our previous computer modeling suggests that G65YCu_BMb is a closer structural model than F33YCu_BMb, for the positioning of Tyr in a similar position as Tyr244 of bovine HCO, and G65YCu_BMb is slightly more active, we choose to focus our efforts on F33YCu_BMb in this study, because it is much more stable and crystallizable than G65YCu_BMb, while still displaying HCO activity.

While this success of biosynthetic models is exciting, they still release 7–32% of reactive oxygen species (ROS) during turnover, in comparison with almost no ROS in native oxidases.^20,21 Release of such ROS is an indicator of subpar efficiency of the enzyme. Successful rational improvement of the enzyme function to eliminate this ROS release would not only be a true test of our understanding of native enzymes, but also allow application of this knowledge to the design of more efficient artificial ORR enzymes or catalysts for use in fuel cell cathodes.

Previously, we have investigated the structural features of the oxy- and peroxo- intermediates of F33Y-Cu_BMb using EPR and crystallographic studies.²¹ Based on the results, we proposed that oxygen activation for complete reduction to water could be attributed to engineered protic residues that organize an extended hydrogen bonding network of water molecules in the active-site. While rational protein design has been extremely successful, it has largely focused on design of interactions directly to the substrate or cofactor. On the other hand, extended hydrogen bonding networks, particularly those mediated by water molecules are well known in natural enzymes;^23–29 however, reports of rationally incorporating such water mediated interactions have been limited.^30–32 Building on our previous results, we herein report incorporation of a Glu residue, which is known to be important in both HCOs and bd oxidases, into Cu_BMb and F33Y-Cu_BMb through rational extension of the water-mediated hydrogen bonding network in its active site, and show that the introduced Glu plays a key role in eliminating ROS formation. In addition, we have obtained insights into the role of Glu using high-resolution crystal structures and EPR of cryoreduced oxy-I107E-Cu_BMb.

MATERIALS AND METHODS

Protein mutagenesis, expression, and purification.

Site directed mutagenesis, and protein expression and purification were carried out as previously reported,²⁰ with minor changes to improve yield and purity. The source of the cloned gene is described earlier;³³ the sequence corresponds to Uniprot P02185–1. For consistency, sequence numbering used here and in the accompanying manuscript omits the initiator methionine, which has been the convention in the literature since the initial sequencing of myoglobin obtained directly from sperm whale tissue revealed that the protein sequence begins with the subsequent valine.³⁴

Expression was carried out with E. coli BL-21 DE3 Star strain (Life Technologies); other strains produced inferior yield and purity. Dialysis of protein after refolding was carried out against 10 mM tris sulfate pH 8.0. Dialysis was limited to two exchanges - first for 4 hours, second overnight. Protein was loaded onto the size exclusion column within 12 hours of completing dialysis. Protein batch activity was validated by oxygen reduction activity assays.^17,20,35

Oxygen reduction assays.

The oxygen reduction rate assays were carried out as previously reported, by first determining the rates of O₂ consumption in the absence and presence of the ROS scavengers, catalase and superoxide dismutase, and then calculate the rates of water and ROS formation using equations described in a previous publication,^17,20,35 using a Clark type oxygen electrode, ascorbate and TMPD as reductant and mediator, respectively, and ROS scavengers.

Oxygen affinity assays.

Measurements of O₂ affinity were performed as described earlier.¹⁵

Crystallization, I107E-Cu_BMb.

Resting state (aquo-met-ferric) crystals of I107E-Cu_BMb were grown under modified conditions previously reported.²⁰ Sitting drop vapor diffusion was utilized to increase drop size and allow for larger crystals. Protein was concentrated to 1.5–2.0 mM in 20 mM tris sulfate pH 8.0 buffer. Crystallization solution was composed of 100 mM tris sulfate pH 8.6, 200 mM sodium acetate, and 30% w/v PEG 10,000. Solutions were mixed 1:1 to yield a final drop volume of 150–200 μL. This was equilibrated by vapor diffusion against 30 mL of crystallization solution, at 4°C, in dark. Crystals formed within 1 week. Two crystal forms were observed - long needles which diverged into fan shapes (P2₁) and thicker rectangular crystals (P2₁2₁2₁). The orthorombic crystal form was propagated by seeding new crystallization drops.

Crystallization, I107E-F33Y-Cu_BMb.

Resting state (aquo-met-ferric) crystals of I107E-Cu_BMb were grown under modified conditions previously reported.²⁰ Sitting drop vapor diffusion was utilized to increase drop size and allow for larger crystals. Protein was concentrated to 1.5–2.0 mM in 20 mM tris sulfate pH 8.0 buffer. Crystallization solution was composed of 100 mM tris sulfate pH 8.6, 200 mM sodium acetate, and 30% w/v PEG 10,000. Solutions were mixed 1:1 to yield a final drop volume of 150–200 μL. This was equilibrated by vapor diffusion against 30 mL of crystallization solution, at 4°C, in dark. Crystals formed within 1 week.

Crystal preparation, oxy-I107E-Cu_BMb.

Oxy-I107E Cu_BMb crystals were prepared by anaerobic reduction of aquo-met-ferric I107ECu_BMb and exposure to oxygen, based on previously reported methods.^36,37 Several large and robust orthorhombic crystals were harvested into a small container. This was brought into an anaerobic atmosphere chamber (Coy Labs) by 27 headspace purges, down to 8 in. Hg of vacuum each, in the antechamber. Subsequently, crystals were transferred to a small finely-fritted filter reservoir, filled with degassed crystallization solution containing 20 mM dithionite. These crystals were left to soak for 10 minutes to reduce protein and eliminate residual oxygen. Soaking solution was gently removed from the reservoir through the frit using a syringe, driven by a syringe pump and connected with plastic tubing to the bottom of the filter reservoir. Solution was replaced by manually adding fresh degassed solution into the filter reservoir. This washing was repeated four times total. Deoxy crystals were harvested into small (~2 μL) drops on slides and sealed in a high pressure bomb. The bomb was removed from the chamber and immediately pressurized to ~1500 psi O₂ for 1–10 minutes. The bomb was carefully vented over 30s and the crystal harvested after brief soak in O₂ supersaturated cryoprotectant (50% PEG 400 supersaturated with O₂ under similar pressurization conditions).

Crystallography, general methods.

Unless otherwise noted, X-ray diffraction data were collected under cryogenic conditions, data reduction and scaling were carried out with HKL2000, refinement was carried out in Phenix, and model building was carried out using Coot. Graphics were generated using VMD with Tachyon rendering. Omit map images generated using PyMol.

Crystallography, met-I107E-Cu_BMb.

X-ray diffraction data were collected on beamline x29 at NSLS. The structure was solved using molecular replacement using the structure of Cu_BMb (PDB ID: 4FWZ) as a model. Sidechains and water molecules were rebuilt manually and iteratively with refinement.

Crystallography, met-I107E-F33Y-Cu_BMb.

X-ray diffraction data were collected on beamline 21-ID-G at APS. The structure was solved by isomorphous replacement using the backbone and heme coordinates from the crystal structure of F33Y-Cu_BMb (PDB ID: 4FWX). R-free flags were copied from the 4FWX dataset and extended to the resolution of this dataset with random selection. Sidechains and water molecules were rebuilt manually and iteratively with refinement.

Crystallography, 1.18Å oxy-I107E-Cu_BMb.

Single crystal spectroscopic data and X-ray diffraction data were collected at SSRL beamline 11–1. The oxygen-bound state of the protein was confirmed using the microspectrophotometer;^38,39 high attenuation of the X-ray beam was used to minimize radiolytic reduction. Data reduction and scaling was carried out with XDS, pointless, aimless, and truncate, automated using the script ‘autoxds.’ The structure was solved by isomorphous replacement using the backbone and heme coordinates from the crystal structure of F33Y-Cu_BMb (PDB ID: 4FWX). R-free flags were copied from the 4FWX dataset and extended to the resolution of this dataset with random selection. Refinement was performed with Phenix and REFMAC. Sidechains and water molecules were rebuilt manually and iteratively with refinement. The O-O group was modeled as peroxide (PER), which would be consistent with the final state in the crystal, due to reduction by the X-ray beam.

Cryoreduced oxy-ferrous EPR sample preparation.

Each protein was concentrated to ~1–3 mM concentration and reduced with excess dithionite anaerobically under nitrogen atmosphere. Dithionite was removed under anaerobic conditions by a small GE Sephadex G-25 desalting column. The protein was reconcentrated anaerobically. The protein solution was removed from the anaerobic chamber and immediately, slowly, and gently sparged with pure oxygen gas from a syringe. The protein was briefly centrifuged to remove bubbles and mixed with pure glycerol to yield 15% final glycerol concentration. This solution was quickly transferred to an EPR tube and slowly frozen with liquid nitrogen.

Samples were transferred to a glass dewar filled with liquid nitrogen and irradiated with ~3 Mrad of gamma radiation over 440 minutes using a Gammacell 220 irradiator containing a ⁶⁰Co source. Irradiated samples were stored and transported in liquid nitrogen at all times.

EPR spectroscopy.

X-band CW EPR spectroscopy was performed using a Bruker ESP-300 spectrometer equipped with an Oxford Instruments ESR-900 helium flow cryostat. EPR spectra were recorded at 20K, a modulation frequency of 100 kHz, a modulation amplitude of 5 G, microwave power of 2 mW and with a sweep rate of 10 G/s. Spectra were recorded at microwave frequencies of approximately 9.364 GHz, with the precise microwave frequencies recorded for individual spectra using a Hewlett-Packard Microwave Frequency Counter (HP5352B).

Safety Statement.

No unexpected or unusually high safety hazards were encountered with the reported work.

RESULTS AND DISCUSSION

To engineer a Glu into the water mediated hydrogen bonding network of Cu_BMb and F33Y-Cu_BMb, we start with their crystal structures (Figure 1, left).²⁰ In our previous study, we proposed that the presence of W1 in the active site of both proteins and its interactions with active site residues are crucial for imparting complete reduction of O₂ to H₂O, instead of partially reduced ROS such as superoxide or peroxide.³⁶ The location and choice of introducing a Glu around the active site were based on visual analysis and basic rotamer modeling in PyMol. The I107E mutation showed the most optimal H-bonding interaction with W1, with minimal perturbation of the active site (Figure 1, center). This choice has been corroborated by a previous study showing that the I107E mutation improved NO reduction activity of a related model, presumably due to improved protonation.⁴⁰

Figure 1.

Structures of Cu_BMb variants. (left) Crystal structures of Cu_BMb and F33Y-Cu_BMb, highlighting water-mediated hydrogen-bonding network in the active site. (center) Computational model of I107E-Cu_BMb based on crystal structure of Cu_BMb and I107E containing Mb variant from PDB 3M38. (right) Crystal structures of I107E-Cu_BMb and I107E/F33Y-Cu_BMb.

Construction, expression, and purification of I107E-Cu_BMb and I107E/F33Y-Cu_BMb was carried out using a protocol reported previously,²⁰ and their identities were confirmed by sequencing of the DNA and mass spectrometry of the purified proteins. As purified, the proteins displayed a Soret peak at 408 nm and a visible spectrum indicative of a high spin ferric heme (Figure S1), consistent with the aquo-ferric (met) resting state of Mb.⁴¹ The extinction coefficients of the Soret peak in the resting states of I107E-Cu_BMb and I107E/F33Y-Cu_BMb were determined to be 181 mM⁻¹ and 175 mM⁻¹, respectively.

The successful incorporation of a carboxylate group to interact with the water-mediated H-bonding network of both proteins was confirmed by obtaining crystal structures of the resting states of I107E-Cu_BMb and I107E/F33Y-Cu_BMb determined at 1.70 Å and 1.47 Å resolution, respectively (Figure 1, right). In both structures, the active site contains a water molecule (W0) coordinating the heme and forming a H-bond with His64 and a second water molecule (W1) near His29 and engaging in a H-bond with W0 (Figure S2), similar to those found in previous structures of Cu_BMb variants, suggesting that introducing the I107E mutation caused almost no perturbation of the active structure, including the delicate H-bonding network. More importantly, the structure revealed that the W1 near His29 also engages in a H-bond with Glu107, as predicted in our initial computational design of the mutants.

Having confirmed the presence of a new H-bond in both mutants, we proceeded to investigate its effect on the functional properties of the enzymes.²⁰ As shown in Figure 2, Cu_BMb reduced O₂ at a total rate of 0.49 s⁻¹, producing H₂O at 0.33 s⁻¹ and ROS at 0.16 s⁻¹, which amounts to 32% selectivity for ROS as reported previously.²⁰ Introducing I107E almost completely eliminated the ROS formation. Similarly, F33YCuBMb reduced O₂ at a total rate of 1.56 s⁻¹, producing water at 1.1 s⁻¹ and ROS at 0.5 s⁻¹, which equates to 29% selectivity for ROS.²⁰ Introducing the I107E mutation into F33Y-Cu_BMb reduced the ROS formation to <5% in I107E/F33Y-Cu_BMb. Therefore, the I107E mutation has resulted in effective elimination of ROS release in both proteins. The rate of water formation of Cu_BMb was roughly doubled by the I107E mutation. The I107E mutation did not increase the rate of water production of F33Y-Cu_BMb.

Figure 2.

Rates of O₂ consumption yielding water (blue) or reactive oxygen species (ROS) (red), measured and calculated as reported previously.^20,17,35 Rates of Cu_BMb, F33Y-Cu_BMb, have been reported previously²⁰ and are included here for visual comparison.

Interestingly, we noted that the raw O₂ consumption traces as a function of O₂ concentration for the I107E mutants showed rate beginning to saturate at lower O₂ concentrations than in the variants without the mutation (Figure S3), suggesting that the I107E mutation imparts enhanced oxygen affinity. We confirmed this suggestion by measuring the oxygen affinities of these variants using laser-flash photolysis (Table S1 and Figure S4), which indeed showed that the I107E mutation resulted in substantial increases in O₂ affinity for both proteins. Previous studies at tuning the O₂-binding affinity of Mb have shown that mutations of Leu29 and Ile107 are critical for tuning the affinity.^42–44 An increased O₂ affinity has been typically associated with larger hydrophobic sidechains and water exclusion in this region. The observation of increase in affinity with the introduction of a charged sidechain is noteworthy and has not been previously reported, to our knowledge. Furthermore, it is known that increased O₂ affinity slows auto-oxidation and thus release of ROS from myoglobin,^45–49 which may contribute to the effective elimination of ROS species formation during the O₂ reduction by the I017E-Cu_BMb and I107E/F33Y-Cu_BMb.

In order to gain further understanding of how the introduced Glu residue tunes the reactivity of Cu_BMb and F33Y-Cu_BMb, we sought to probe the interactions of the oxygen-bound intermediate. We and others have previously shown that EPR is a valuable tool for probing the interactions within the active sites of heme and non-heme iron proteins.^50–68 In particular, cryoreduction of oxy-ferrous heme proteins to yield EPR-active peroxo- and hydroperoxo-ferric heme states, which retain the conformation of the oxy heme precursor, has been extremely successful in probing the interactions of oxygen with its active-site environment in oxygen-activating proteins.^50–58,65 As mentioned earlier, this technique was used to understand the structural features responsible for F33Y-Cu_BMb reactivity.³⁶ Here, we apply this technique to compare the structural features and interactions of the newly designed I107E-Cu_BMb and I107E/F33Y-Cu_BMb.

The EPR spectra of 77K cryoreduced oxy-Cu_BMb and oxy-F33Y-Cu_BMb, and their I107E variants are shown in Figure 3 and their g-values are summarized in Table 1. Both cryoreduced Cu_BMb and F33Y-Cu_BMb spectra collected at 20K exhibit two new species. The EPR parameters of the first species, with g₁ = 2.24 in both proteins, suggest that it is a ferric peroxo state in which the peroxo ligand engages in a hydrogen bond with the distal His64.^50–53,61 The somewhat broadened g₁ feature in F33Y-Cu_BMb indicates a slight conformational distribution. The second species with g₁ = 2.29 in both proteins is in the range characteristic of ferric-hydroperoxo intermediates.

Figure 3.

Representative 20 K EPR spectra of oxy-Cu_BMb proteins after 77K radiolytic reduction with a ~3Mrad dose of γ radiation from ⁶⁰Co, and after subsequent stepwise annealing as indicated (sharp signal marked by asterisk is due to hydrogen atoms radiolytically generated at 77K in quartz EPR tube).

Table 1.

g-Values of Cryoreduced Oxy-ferrous Myoglobins

Protein	g1	g2	g3
CuBMb
peroxo	2.24	2.13	1.96
hydroperoxo1	2.29	2.16	1.94
hydroperoxo2	2.34	2.18	1.93
F33Y-CuBMb
peroxo	2.24	2.13	1.96
hydroperoxo1	2.29	2.17	1.95
hydroperoxo2	2.34	2.19	1.94
I107E-CuBMb
peroxo	2.27	2.16	nd
hydroperoxo A1	2.29	2.20	nd
hydroperoxo B1	2.32	2.17	1.94
I107E/F33Y-CuBMb
peroxo A	2.23	2.12	nd
peroxo B	2.27	2.16	nd
hydroperoxo A1	2.28	2.17	nd
hydroperoxo B1	2.32	2.17	1.95
hydroperoxo C2	2.34	2.19	1.91

¹Primary hydroperoxo product formed at 77K.

²Hydroperoxo product formed upon annealing.

Annealing of cryoradiolytically generated peroxo-heme species has been shown to provide further insights into the local environment. Therefore, we performed EPR of the frozen samples after sequential annealing at progressively higher temperatures. Spectroscopic traces of the samples at key points are shown in Figure 3, and the full series of spectra is shown in Figure S5. Annealing of Cu_BMb and F33Y-Cu_BMb to 145K causes no change in speciation and only slight changes in lineshapes reflecting conformational changes. Annealing the Cu_BMb peroxo species for 15 minutes at 199K generates a hydroperoxo state with g₁ = 2.34, which can be assigned as arising from protonation of some g₁ = 2.24 peroxo centers and some rearrangement of the conformation of the g₁ = 2.29 hydroperoxo state into a more rhombic hydroperoxo state. In contrast, annealing of cryoreduced F33Y-Cu_BMb for only 6 minutes at 199K causes essentially complete protonation of the peroxo form with conversion to the g₁ = 2.34 hydroperoxo state, consistent with facilitated proton transfer by this mutation.

The 77K EPR spectra of the cryoreduced oxy-I107E-Cu_BMb variant are distinctly different from those of the oxy-Cu_BMb and its F33Y variant. The major signal produced at 77K, with g₁ = 2.27 is assigned as a peroxo ferric species significantly more rhombic than the peroxo species in the other two proteins. This species is also observed in the 77K of the double mutant. The higher rhombicity of the ferric peroxo signals of these I107E-containing mutants compared to those without the mutation is an indicator of a stronger H-bond interaction between the cryogenerated peroxo ligand, and thus the parent O₂ ligand in the ferrous-oxy complex, and a nearby hydrogen bond donor. Furthermore, prior to annealing, a highly rhombic hydroperoxo signal with g₁ = 2.32 is observed in the I107E mutants which is not observed in Cu_BMb and F33Y-Cu_BMb until they were annealed at 173K (Figure S5). The lower rhombicity hydroperoxo ferric signal, g₁ = 2.28-2.29, manifests as a shoulder of the major peroxo peak in both I107E variants. The spectroscopic distinction between the detected hydroperoxo intermediates with g₁ = 2.29 and 2.32-4 may be explained by different geometry of Fe-O-O-H unit.⁵⁷

In the case of cryoreduced I107E mutant we observed the nearly complete decay of the major peroxo signals in the I107E mutant upon annealing at 180K for one minute (Figure 3 and S5). The decay of the peroxo signal is concomitant with the increase of the g₁ = 2.32-2.34 hydroperoxo signal, indicative that protonation of the peroxo species has produced the hydroperoxo state. This observation of rapid protonation of the major peroxo species in 1 minute annealing at 180K to yield hydroperoxo in the protein containing the I107E mutation is in stark contrast with the results of annealing of Cu_BMb, without the I107E mutation, where residual peroxo state remains after three minutes of 180K annealing, and strongly supports the generation of a very mobile proton in the active site by this mutation. Intriguingly, in the double mutant, the presence of the F33Y change slightly stabilize the peroxo state and slowed its protonation at 180K (Figure 3 and S5).

Further analysis of the time-dependence of the final annealing spectra in Figure S5 is shown in Figure 4. While all samples eventually decay to an EPR silent state, the proteins lacking the I107E mutation require longer times and higher temperatures (199K) than the proteins containing the mutation (180K). Consistent with previous results on myoglobin, subsequent cryoreduction of this EPR-silent state in I107E/F33Y-Cu_BMb produces a g₁ = 2.54 signal consistent with reductive generation of a ferric hydroxo (Fe(III)-OH; Figure S6). These results confirm that (1) the EPR-silent state is the ferryl compound II (Fe(IV)=O), and furthermore that (2) the active site environment of I107E variants of Cu_BMb facilitates protonation of the Fe ligand.

Figure 4.

Temperature dependence of the g₁ peak intensity (in arbitrary units) of the peroxo (gray ■, □) and hydroperoxo (green ● and purple ▽, ▼) states, showing faster decay at lower temperatures in I107E mutants.

This contrast between the decay kinetics of the (hydro)peroxo species of I107E and parent enzymes at cryogenic temperatures further supports the design hypothesis, that Glu107 stabilizes an additional proton in the active site, enabling the hydrogen bonding network to efficiently activate oxygen by quickly delivering two protons to promote heterolytic O-O bond cleavage during turnover. This study provides structural and spectroscopic basis for previously observed D₂O/H₂O isotopic effects in electro-chemical studies on the role of I107E mutation on the rate-determining step of O₂ reduction in the Cu_BMb variant,¹³ and demonstrated the dominant role of proton transfer residue in HCO activity.

In order to gain further insight into the structural basis for oxygen activation of these efficient enzymes, we obtained the crystal structure of the dioxygen-bound intermediate. Oxy-I107E-Cu_BMb crystals were obtained as previously reported, by reducing ferric crystals, followed by rapid exposure to oxygen (details in SI). The active site structure of oxy-I107E-Cu_BMb, solved at 1.18 Å resolution, is shown in Figure 5. While the occupancy of the dioxygen molecule is incomplete (see Figure S7 for alternative occupancies), likely due to large crystals and short soak times to prevent autooxidation, the structure still provides a structural basis for understanding the role of Glu107. As predicted from design, a water molecule is present within hydrogen bonding distance of the bound dioxygen, and furthermore, this water molecule engages in H-bonds with Glu107 and His29. The geometry of the H-bonding interactions with oxygen is close to ideal, suggesting that the interactions likely polarize the O₂ bond and could facilitate reduction and O-O bond cleavage.

Figure 5.

Structure of the active site of oxy-I107E-Cu_BMb at 1.18 Å resolution, with pertinent interatomic distances labeled. Distal pocket omit map shown in Figure S2.

CONCLUSION

Taken together, the results above demonstrate that the extension of the hydrogen bonding network within the active site of our designed biosynthetic models of HCO by introducing Glu107 improves O₂-binding affinity, facilitates protonation of reaction intermediates and eliminates ROS release, resulting in cleaner oxygen reductase chemistry. We have provided X-ray structural and EPR spectroscopic evidence that the enhanced catalytic performance is a result of Glu107 engaging in strong hydrogen bonds with the active site water network, stabilizing protons in the active site (Scheme 2), which are then available for efficient oxygen polarization, activation, and protonation of the formed peroxo- or hydroperoxo intermediates during turnover. These results support the role of the D-pathway-terminating Glu residue, and potentially the Glu residue near the heme d in bd oxidases, in providing protons to the bound dioxygen during the first steps of oxygen reduction. Moreover, the results presented here, extensively characterizing the hydrogen bonding networks of 4 different variants of Mb, clearly demonstrate that designing long-range, water-mediated hydrogen-bonding networks - a feature that has seen some but, minimal pursuit in either synthetic models using organic molecules as ligand or rational enzyme design strategies^30–32 - can play a key role in conferring and fine-tuning enzymatic function, and should be pursued more actively in biomimetic modeling of native enzymes.

Scheme 2.

Proposed interaction of Glu107

ABBREVIATIONS

Mb

myoglobin

swMb

sperm whale Mb

WTMb

wild type swMb

Cu_BMb

L29H F43H swMb

HCO

heme-copper oxidase

EPR

electron paramagnetic resonance

ENDOR

electron-nuclear double resonance